Jackson Kontny·Nov 3Prefix sums, from O(1) range queries to parallel scansOne row of running totals answers any range-sum question with a single subtraction. Building that row looks stubbornly sequential, yet SIMD…
Matteo Ferrabone·Nov 2Pillow-SIMD, everywhere: bringing fast image processing to ARMHow I took Pillow-SIMD to Pillow 13, wrote NEON kernels for arm64, and made Python’s favourite imaging library up to six times faster than…
Nehal Singhal·Nov 1PyTorch to AVX2: MLIR Compiler Pt. 3 | Memory BufferizationWe’re now and the most optimized version of our MLIR code possible without switching from the tensor dialect to a ‘memref’ dialect of real…
Nehal Singhal·Oct 26PyTorch to AVX2: MLIR Compiler | Generalization, Tiling & FusionFor some background, we’re now at the point where we have a .MLIR file consisting of the func, linalg, tensor and arith dialects. Our goal…A response icon1A response icon1
Rahul Kulkarni·Aug 25Solana is about to make inefficient transactions expensive.been seeing a lot of discussion around solana’s SIMD-0553 lately.
Zahran Yahia Khan·Aug 22Why More SIMD and More Threads Made My C++ Engine SlowerWhat I learned optimizing a 500-million-user bitmap query engine to the DDR4 memory-bandwidth ceiling
InOrdinary Programmerbyhuizhou92·Aug 14Go 1.27 SIMD Benchmarked — Lookup Drops From 76 ns to 14.5 ns, and the Portable Package Costs 2.3xTwo comparisons on one M5 — 5x between scalar code and SIMD, another 2.3x between archsimd and the portable SIMD package
Rajat Jain·Aug 3How I Sped Up OpenSearch Multi-Valued Aggregations by 439x Using Roaring BitmapsA deep dive into custom Lucene codecs, SIMD bitset vectorization, and benchmarking on 1,000,000 Wikipedia articles.
Amit Shukla·Jul 5Python Wrote the Model. Mojo 1.0 🔥Made It Fly.🎥 I’ve also turned this into a full, free video series — 16 chapters taking you from “what is Mojo?” to ownership, pointers, and the…
Karan·Jun 13The Point Where One Machine Is Never EnoughContinuing the series where I share what I’m learning about cloud computing each week.
